A smartphone app to perform spirometry
Many of us would love to have our patients monitor FEV1 rather than peak flow for monitoring a number of lung diseases. Unfortunately, the equipment required is typically too expensive and difficult to use outside the clinical setting.
I have seen several apps for various mobile devices, mostly the iPhone, that purport to allow them to be used as aspirometer. At least one of these requires additional hardware that connects to the smartphone. The others use the microphone to attempt to determine air volume andflow. I was not surprised that the latter category is offered for entertainment purposes and not clinical use. They are completely inaccurate and work poorly.
I was therefore surprised to see another offering that again uses only the mobile device microphone. Unlike previous offerings, however, the new app, SpiroSmart, is the result of a research and development effort at the University of Washington. The engineers collected data from 52 volunteers using an iPhone and during the same session performing formal spirometry.
With the iPhone, they calculated the exhaled flow rate by estimating models of the user’s vocal tract and the reverberation of sound around the user’s head. They also determined the impact of the distance of the phone from the participant’s mouth and were able to adjust for this. The app includes user coaching using visual cues.
The modeling process is quite complex and fairly innovative. The designers further refined their algorithm using machine learning techniques with the actual spirometry data. Much of the actual waveform processing is too complex for current smartphones. The bulk of this processing needs to be done on a larger platform in “the cloud.” This is not unlike other compute intensive appscurrently available.
The result is an app that performs a very complex analysis of the sound received by the device microphone during a forced expiratory maneuver. Using this technique, it is possible to measure the FVC, FEV1 and PEF in test subjects to within 5% of the formal spirometric values andto personalize the algorithm for individual subjects,improving the accuracy slightly more.
The 52 test subjects mostly had normal lung function and the few with airflow obstruction showed less reliable results, although these were improved to within acceptable levels with the personalization procedure. Most of the error occurs in the measurement of FVC as the audio signal becomes undetectable at lower flows. The FEV1 measurement appears to be fairly reliable. Not surprisingly, the application must be used in a very quiet environment since extraneous noise causes problems.
The SpiroSmart engineers are in the process of investigating how well this application does detecting changes in lung function over time, as with an asthma exacerbation. No information about this is yet available.SpiroSmart is not intended as a substitute for clinical spirometry, but rather as a home-based solution that can increase compliance and monitoring through the convenience of a mobile phone. The authors are hopeful that it will be useful and cost effective in this setting.
Although the article describing SpiroSmart does not specifically say, it would seem that they should be able to offer this on multiple mobile platforms.